OBSERVED IN REAL USE · Anthropic 2026
3%
of this role’s work is already showing up in real Claude usage (Anthropic Economic Index).
The robots can help. They can’t replace the room you read.
28% of this role’s O*NET tasks are within reach of today’s AI. That is the core-weighted exposure score from Eloundou et al. 2023 (“GPTs are GPTs”). It measures a capability ceiling, not a headcount forecast. Mostly safe. AI helps around the edges, but the job stays human.
WHAT AI CAN ALREADY DO
- Complete administrative paperwork or supervise the preparation or maintenance of records, forms, or reports
- Set up employee work schedules
- Convey correctional officers' or inmates' complaints to superiors
- Maintain knowledge of, comply with, and enforce all institutional policies, rules, procedures, and regulations
- Conduct evaluations of employees' performance
WHAT IT STILL CAN’T
- Maintain order, discipline, and security within assigned areas in accordance with relevant rules, regulations, policies, and laws
- Respond to emergencies, such as escapes
- Supervise and direct the work of correctional officers to ensure the safe custody, discipline, and welfare of inmates
- Supervise or perform searches of inmates or their quarters to locate contraband items
- Monitor behavior of subordinates to ensure alert, courteous, and professional behavior toward inmates, parolees, fellow employees, visitors, and the public
THE HONEST PART. A percentage is not a pink slip. High exposure usually means a role shrinks and shifts toward judgment, direction and responsibility: the parts a model can’t sign its name to. Exposure ≠ displacement. Breathe.
"My job is 28% cooked. What’s yours?"
SOURCES: O*NET 30.3 occupational tasks · Eloundou et al. 2023 (“GPTs are GPTs”,
arXiv:2303.10130) · Anthropic Economic Index 2026 (CC-BY) |
how this is calculated |
last updated 2026-07-16